The Agent Skills Directory

[PROMPT_INJECTION]: The skill reads and processes content from repository markdown files, creating a potential surface for indirect prompt injection. Malicious instructions placed in these files could theoretically influence agent behavior. However, the risk is mitigated by the skill's strict reliance on direct tool evidence and mandatory human review before any file modifications.
[COMMAND_EXECUTION]: Local shell commands including git grep and git ls-files are executed through an included script to identify duplication. These operations are restricted to the local repository and follow safe coding practices to prevent argument injection.
[EXTERNAL_DOWNLOADS]: The skill references external URLs for documentation and best practices from trusted organizations like Anthropic and Microsoft. These are used for informational purposes only and do not involve downloading or executing external code.

extract-ssot